Every recognized target with no qualifying signal right now — scored, clustered by demand family, and filed with the trigger that would make you move within 48 hours.
No company here has a live, Bay-eligible, rubric-qualifying role this cycle. Nine from prior dormant lists — Stripe, Abridge, Glean, Vanta, Render, Socure, Kin, Suno, and Salesforce — now carry active signals and have been pulled for the next active-signal cycle. Adobe stays outside this list until its current design inventory resolves.
Two companies — Ramp (#90) and Databricks (#20) — have live signals that are geographically ineligible. They stay here because routing them to the active list would create false urgency for roles you can't take. Their triggers are geographic: a Bay-eligible or US-remote copy of the existing role.
What's left is organized by the design problem each company would hire you to solve. The clusters are demand families — when a signal fires at any company below, the cluster tells you which portfolio proof to reach for before you even open the entry.
Scan trigger conditions weekly against your LinkedIn alerts and ATS notifications. When a trigger fires, score the actual role against the role rubric before acting. The entry gives you the company score, portfolio match, and tier archetype — read the corresponding tier playbook before outreach. When using any portfolio match in outreach, apply BCG DV attribution per the standard rule. Cummins/ZED Connect appears frequently as a portfolio match in Clusters 3, 7, 8, and 10 — in outreach, it supports rather than leads. Skip the research phase entirely.
Company scores are carried forward from prior published assessments. No role score is assigned because no posting exists — when a trigger fires, the role rubric scores the actual posting. A universal trigger applies to every entry: a fresh, Bay-eligible or US-remote Staff+ or Director+ product design posting, a new design executive inside 30 days, or a fresh funding event paired with design hiring. Each entry adds a demand-type trigger specific to that company's design problem.
1. Agent Trust and Supervision
Autonomous agents handling full conversations, executing integrations, processing documents. Every company in this cluster would hire you to solve the same problem: how does a human inspect and correct agent behavior before acting on its output? Have Brand Pulse ready (agent-performance visibility), Carrier IQ (recommendation provenance in a live workflow), and your TinyFish production experience with agent traces.
1. Decagon. AI customer-support agents that handle full conversations autonomously. Series B, SF. Midstream agent platform. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Agents that resolve full conversations autonomously rather than deflecting to humans.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Their ops teams need to see why an agent said what it said before deciding whether to intervene. Brand Pulse's monitoring architecture answers this directly. Alibaba's +20% daily transactions shows you've designed visibility at platform scale.
- Verdict: Ignore-until-trigger. Demand-type: agent configuration, evaluation, escalation, or operational-visibility design.
2. Cresta. Real-time AI coaching for contact-center agents during live conversations. Series D, SF. Midstream. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Coaching that happens during the call, not after — real-time recommendation while the agent is still talking.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cresta's agents recommend actions during a live call, and the human needs to see why before following the suggestion. Carrier IQ's quote-automation provenance is the exact match. TinyFish agent-trace experience gives you the operational layer a portfolio-only candidate doesn't have.
- Verdict: Ignore-until-trigger. Demand-type: human-agent coaching, recommendation explanation, or real-time intervention design.
3. Retool. Internal-tool builder with AI agent capabilities for enterprise. Series D (~$3.85B valuation), SF. Paco Viñoly leads design. Midstream platform. Growth-stage platform. C9/12 (AI 2, Stage 3, Design 2, Traj 2).
- Best at: Letting non-engineers build production internal tools — now extending that model to agent builders.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Retool's expanding agent-builder surface needs governance design — permissions, audit trails, deployment controls. Your TinyFish governance experience covers this. Trust essay's reversibility framework addresses the next question: what happens when a customer-built agent breaks something.
- Verdict: Ignore-until-trigger. Demand-type: agent-builder governance, auditability, permissions, or deployment-review design.
4. Workato. Enterprise integration and automation platform with autonomous workflow agents. Late-stage, Mountain View. Enterprise platform. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Breadth of enterprise integration connectors (1,000+) — the automation surface is wider than any competitor.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Workato's hardest problem: an autonomous integration modifies data across three systems and one modification was wrong. Trust essay's reversibility framework (Watch → Verify → Delegate) addresses this directly. Red Cross's 6-systems-to-1 consolidation shows you've designed recovery across interconnected systems.
- Verdict: Ignore-until-trigger. Demand-type: agentic automation, exception recovery, or cross-system rollback design.
5. Instabase. AI-powered document processing and extraction for enterprise. Series C, SF/NYC. Midstream. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Unstructured document processing with enterprise-grade extraction accuracy across document types.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Instabase's extraction-review workflow is the Output Review pattern from your Trust essay: human inspects model confidence before releasing a decision. Thermo Fisher's 100% adoption in pharma shows you can design review workflows that regulated users actually complete.
- Verdict: Ignore-until-trigger. Demand-type: review queues, extraction-confidence surfaces, provenance, or workflow-release design.
6. Gong. Revenue intelligence platform using AI to analyze sales conversations. Late-stage (~IPO-track), SF/Palo Alto. Enterprise platform. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Turning conversation data into revenue predictions — the category creator for conversation intelligence.
- Signals: No recent signal. Last assessed Issue #10.
- Match: A manager sees an AI-generated deal-risk flag and needs to understand the evidence before intervening. Carrier IQ's recommendation-provenance design fits here. Alibaba's NPS improvement is your evidence for making complex data actionable at scale.
- Verdict: Ignore-until-trigger. Demand-type: manager-agent intervention, recommendation explanation, or deal-intelligence design.
7. Alation. Data intelligence and governance platform with AI-powered discovery. Late-stage, Redwood City. Enterprise platform. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Data catalog governance — the company that defined the category and still leads in enterprise adoption.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Alation's core challenge is AI search that respects governance without making discovery feel like a compliance exercise. Your Trust essay's permission-aware retrieval framework addresses this. Alibaba's search redesign (+20% transactions) is your evidence for discovery at enterprise scale.
- Verdict: Ignore-until-trigger. Demand-type: AI search, policy controls, permission-aware retrieval, or data-governance UX.
8. Hyperscience. Document-processing automation with human-in-the-loop correction. Growth-stage, NYC. AI-native. C7/12 (AI 3, Stage 1, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Human-in-the-loop correction that actually improves model accuracy over time — the feedback loop is the product.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Hyperscience's central interaction is the Correction-to-Next-Run pattern from your Trust essay: an operator corrects an extraction error and the system learns from it. Your Human-Agent System Design artifact extends this.
- Verdict: Ignore-until-trigger. Demand-type: source-evidence inspection, model correction, or exception-review design.
9. Clari. Revenue operations platform using AI to generate and explain forecasts. Late-stage, Sunnyvale. Enterprise platform. C6/12 (AI 2, Stage 1, Design 1, Traj 2).
- Best at: Revenue forecasting from pipeline data — the platform that RevOps teams use to call the quarter.
- Signals: No recent signal. Last assessed Issue #10.
- Match: A revenue leader sees a forecast change and needs to understand which data drove it and the model's confidence level. Your Inference-Aware UX artifact addresses this directly. Alibaba's platform-scale data visualization supports.
- Verdict: Ignore-until-trigger. Demand-type: forecast explanation, agentic revenue operations, or AI-confidence surfaces.
10. Notion. Workspace platform with AI assistants and autonomous Workers. Series C (~$10B valuation), SF. Downstream application expanding into midstream agent platform. Growth-stage platform. C9/12 (AI 2, Stage 3, Design 2, Traj 2).
- Best at: Flexible workspace that adapts to how teams actually work — now extending that flexibility to autonomous agents (Workers).
- Signals: No recent signal. Last assessed Issue #8.
- Match: What happens when an agent modifies a page that three humans are also editing? That's the Workers problem, and your Trust essay's shared-workspace autonomy framework applies directly. Alibaba's cross-functional sprint methodology shows you can redesign a collaborative platform without breaking existing workflows.
- Verdict: Ignore-until-trigger. Demand-type: Workers, custom agents, permissions, shared-workspace autonomy, or agent-attribution design.
2. Enterprise Coherence and Design Systems
Multiple users, teams, and agents modifying the same artifacts — code, designs, data, documents. The design problem across this cluster is coherence: how do you maintain it when many actors change the same system simultaneously? Pull Alibaba (coherence across a massive B2B platform), Trust essay (shared-workspace autonomy), and the Agent Infrastructure as UX artifact.
11. Figma. Collaborative design platform with AI-powered generation and code translation. Late-stage pre-IPO, SF. Midstream platform. Growth-stage platform. C10/12 (AI 2, Stage 3, Design 3, Traj 2).
- Best at: Real-time multiplayer design collaboration — the tool that made design a team sport. Now the question is whether AI generation fits that collaborative model.
- Signals: No qualifying Staff+/Director+ product design posting found in current inventory. Several unlevelled Product Designer and Design Manager roles present.
- Match: Figma's AI generation challenge is an Intent-Based Interaction problem: a designer says "make this responsive" and needs to steer the result without starting over. Equinox+'s 0→MVP in 3 months shows craft-intensive product speed.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ AI-model interaction, design-to-code round-tripping, or generated-artifact review.
12. Atlassian. Collaboration platform (Jira, Confluence) with Rovo AI agents. Public, SF/Sydney. CDO is Charlie Sutton. Enterprise platform. C9/12 (AI 2, Stage 2, Design 3, Traj 2).
- Best at: Work-management platform adoption at enterprise scale — Jira and Confluence are category defaults. Rovo is their bet on agents that operate across both.
- Signals: Current qualifying-adjacent role is Senior Product Designer, Design Systems — below Staff/Director threshold. No Director+ or Staff+ product design posting found.
- Match: Rovo's specific problem is the permission-aware agent framework from your Trust essay: an agent that reads Confluence and modifies Jira needs clear boundaries visible to the user. Alibaba's cross-functional platform redesign shows coherence across a multi-product surface.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ Rovo, AI-native design systems, or human-agent workflow coherence.
13. Airtable. Structured-data platform with AI-powered automation and Interfaces. Late-stage, SF. Growth-stage platform. C8/12 (AI 2, Stage 2, Design 2, Traj 2).
- Best at: Making structured data accessible to non-technical teams — the spreadsheet-database hybrid that business teams actually adopt.
- Signals: No Staff+/Director+ product design posting found in current inventory. Last assessed Issue #8.
- Match: Airtable's emerging problem fits your Output Review pattern: AI-generated bulk changes across structured records need preview and approval before they execute. Allē's dual-surface redesign (30M members, 3.2× redemption) shows you can redesign a platform with active users without breaking workflows.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ Interfaces, automation preview, or enterprise-administration design.
14. Vercel. Frontend deployment platform with AI-assisted development. Series D, SF. Growth-stage platform. C8/12 (AI 2, Stage 2, Design 2, Traj 2).
- Best at: Developer experience for frontend deployment — the platform that made "ship to preview" feel instant.
- Signals: No recent signal. Last assessed Issue #8.
- Match: Vercel's problem is Agent Infrastructure as UX: understanding and reversing agent-generated changes across code, preview, build, and deployment stages. TinyFish's 0→1 agentic platform build shows you've shipped in this territory.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ agent/developer workflow, deployment-review, or AI-generated code-change design.
15. Snowflake. Cloud data platform with Cortex AI layer. Public, SF/Bozeman. Enterprise platform. C6/12 (AI 1, Stage 2, Design 1, Traj 2).
- Best at: Separating storage from compute for data warehousing — now extending that architecture to AI workloads via Cortex.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cortex needs AI-generated data transformations with audit trails that satisfy both data engineers and compliance teams. Your Trust essay's auditability framework addresses this. Thermo Fisher's 100% adoption in pharma shows governance that regulated users actually trust.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ Cortex AI, governance, or data-observability design.
16. MongoDB. Developer data platform with Atlas AI capabilities. Public, NYC. Enterprise platform. C6/12 (AI 1, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Document-model database that developers choose by default — Atlas is the managed cloud layer.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Atlas AI's challenge is an Inference-Aware UX problem: developers need to understand why a vector search returned what it returned. TinyFish production experience with retrieval and agent architecture gives you the operational layer.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ Atlas AI, retrieval infrastructure, or developer-facing AI design.
17. Cloudflare. Network infrastructure with Workers AI and agent-gateway capabilities. Public, SF. Enterprise platform. C6/12 (AI 1, Stage 2, Design 1, Traj 2).
- Best at: Edge network performance and security at global scale — now positioning as the infrastructure layer agents traverse.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cloudflare's emerging problem is agent identity at the edge: agents acting through the network need identity, permissions, payment authorization, and revocation. Your Trust essay's agent-identity framework addresses this. TinyFish governance experience with agent permissions in production is the practitioner proof.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ agent identity, authorization, or agent-gateway design.
18. Elastic. Search and observability platform with AI-powered security analytics. Public, SF/distributed. Enterprise platform. C6/12 (AI 1, Stage 2, Design 1, Traj 2).
- Best at: Open-source search at enterprise scale — the Elasticsearch ecosystem is the default for log analytics and security observability.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Elastic's security problem: an AI escalates an alert, and the analyst needs to see the evidence chain before acting. Your Trust essay's escalation framework covers this. Red Cross's high-consequence disbursement platform ($847K) shows you can design decision surfaces where acting on bad information has real costs.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ agentic SOC, AI-security alerting, or observability design.
19. Miro. Visual collaboration platform with AI-powered canvas features. Late-stage, SF/Amsterdam. Growth-stage platform. C9/12 (AI 2, Stage 3, Design 2, Traj 2).
- Best at: Infinite-canvas collaboration for distributed teams — the whiteboard that replaced the conference room.
- Signals: No recent signal. Last assessed Issue #8.
- Match: Miro's challenge: AI that synthesizes across a shared canvas must preserve the divergent thinking it's summarizing. Your Human-Agent System Design artifact addresses this directly. Alibaba's cross-functional sprint methodology (3 sprints, multiple stakeholder groups) shows you can design collaborative workflows that respect different users' contributions.
- Verdict: Ignore-until-trigger. Demand-type: refreshed US-eligible Director+ or Staff+ Canvas AI role.
20. Databricks. Unified data and AI platform. Pre-IPO (~$62B valuation), SF/NYC. Enterprise platform. C10/12 (AI 2, Stage 3, Design 2, Traj 3). Current Staff AI role is NYC-based — geographic mismatch. Held here rather than on the active list because the existing role is not Bay-eligible.
- Best at: Unifying data engineering, analytics, and AI on a single lakehouse platform — the company that made "lakehouse" a category.
- Signals: Staff Product Designer, AI Products live in NYC. Senior Product Designer, AI/BI live in Mountain View/SF/Seattle. Neither is Bay-eligible at Staff+ level.
- Match: Databricks' AI/BI problem is pre-execution review: AI-generated data transformations need human inspection before they execute against production data. Your Trust essay's framework addresses this. Thermo Fisher's $20M+ margin platform shows review workflows in high-stakes data environments.
- Verdict: Ignore-until-trigger. Demand-type: Staff+ AI/BI or governance role explicitly open in Bay Area or remote-US.
3. High-Consequence Decision Workflows
Irreversible physical or operational consequences — defense, public safety, autonomous systems. All of them need approval surfaces, evidence inspection, and human override when a wrong action costs lives or missions. Reach for Red Cross (high-consequence disbursement), Trust essay (delegation ladder), and the Human-Agent System Design artifact. Where Cummins/ZED Connect appears below, it provides domain credibility — lead with Red Cross or Trust essay in outreach and use Cummins as supporting proof.
21. Palantir. Operational data platform for government and enterprise decision-making. Public, Denver/Palo Alto. Enterprise platform. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Making massive, heterogeneous datasets operationally useful for analysts and decision-makers — the platform governments actually deploy.
- Signals: No recent signal. Last assessed Issue #8.
- Match: Palantir's users make consequential decisions based on AI-synthesized evidence. The design problem is making that evidence inspectable under time pressure. Red Cross ($847K disbursed, 6 systems → 1, national deployment) is the direct proof.
- Verdict: Ignore-until-trigger. Demand-type: approval surfaces, operational-decision design, or AI-evidence inspection.
22. Applied Intuition. Simulation and development platform for autonomous vehicles. Series D (~$6B valuation), SF. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Full-stack simulation for autonomous driving — the development environment AV companies use to test before road deployment.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Their core problem is operator confidence in simulated autonomous behavior — your Trust essay's confidence framework addresses this. Cummins/ZED Connect (dual-surface fleet management, predictive asset management) adds domain credibility as supporting proof.
- Verdict: Ignore-until-trigger. Demand-type: operator-confidence surfaces, simulation-review, or autonomy-supervision design.
23. Anduril. Defense technology — autonomous systems, sensor fusion, command and control. Late-stage (~$14B valuation), Costa Mesa. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Software-defined defense hardware — the company that brought Silicon Valley product development to military systems.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Anduril's hardest design problem: mission decisions with physical consequences that cannot be undone. Your Trust essay's irreversibility framework applies directly. Red Cross's national-scale high-consequence platform is the proof.
- Verdict: Ignore-until-trigger. Demand-type: mission-decision surfaces, human override, or autonomous-system command design.
24. Vannevar Labs. AI-powered intelligence analysis for defense and national security. Series B, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: AI that augments intelligence analysts rather than replacing them — the human-agent coordination model for classified work.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Vannevar's core interaction is an analyst and an AI working the same intelligence problem, where the design must make clear who owns which part of the analysis. Your Human-Agent System Design artifact fits. TinyFish agent-trace experience adds the operational layer.
- Verdict: Ignore-until-trigger. Demand-type: analyst-agent coordination, intelligence-workflow, or evidence-synthesis design.
25. Primer AI. AI platform for document analysis and intelligence briefing generation. Series C, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Automated intelligence briefings from unstructured document corpora — turning thousands of documents into actionable summaries.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Primer's problem: an AI-generated briefing cites sources, and the reader needs to inspect and challenge those citations before the briefing moves downstream. Trust essay's source-provenance framework addresses this. Carrier IQ's provenance design is the portfolio proof.
- Verdict: Ignore-until-trigger. Demand-type: source provenance, briefing-approval, or citation-inspection design.
26. Peregrine Technologies. AI platform for public-safety case management and operations. Growth-stage, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Case-management intelligence that connects across investigation, decision, and field-operations stages for public safety.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Peregrine needs evidence continuity as a case moves through investigation, decision, and action stages. Red Cross's multi-system consolidation (6 → 1) is the proof you've done this. Trust essay's evidence-chain framework extends it.
- Verdict: Ignore-until-trigger. Demand-type: evidence continuity, case-decision, or public-safety operational design.
27. Shield AI. Autonomous aircraft systems for defense. Late-stage (~$5B valuation), San Diego. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay eligibility.
- Best at: Autonomous flight in GPS-denied environments — the AI pilot that operates where human pilots can't.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Shield AI's command problem: at what confidence level does a human delegate flight control to the autonomous system, and how do they take it back? Your Trust essay's delegation ladder (Watch → Verify → Delegate) maps directly to this. Cummins/ZED Connect supports with dual-surface operational design.
- Verdict: Ignore-until-trigger. Demand-type: human override, autonomous-system command, or mission-control design.
28. Flock Safety. AI-powered public-safety platform — license plate readers, real-time alerts, investigative tools. Series E, Atlanta. AI-native. C7/12 (AI 3, Stage 2, Design 1, Traj 1). Flag: verify Bay/remote eligibility.
- Best at: Objective public-safety evidence at scale — the LPR network that law enforcement agencies actually deploy.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Flock's problem: an alert fires, and the investigator needs to understand system confidence and supporting evidence before acting. Your Trust essay's confidence framework fits. Red Cross's high-consequence decision design is the proof.
- Verdict: Ignore-until-trigger. Demand-type: alert confidence, investigative-evidence, or public-sector policy-control design.
29. Saildrone. Autonomous ocean drones for maritime data collection and defense. Growth-stage, Alameda. AI-native. C7/12 (AI 3, Stage 1, Design 1, Traj 2).
- Best at: Persistent autonomous ocean presence — uncrewed surface vehicles that collect data for months without returning to port.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Trust essay's recovery framework addresses what happens when an autonomous drone encounters an exception mid-mission. Cummins/ZED Connect (fleet management, predictive asset management) supports for remote fleet operations.
- Verdict: Ignore-until-trigger. Demand-type: remote mission-control, fleet-exception recovery, or autonomous-operations design.
30. Astranis. Small geostationary communications satellites. Growth-stage, SF. Hardware-software integration. C7/12 (AI 1, Stage 2, Design 2, Traj 2).
- Best at: Small, affordable geostationary satellites — bringing broadband to markets that can't justify a full-size satellite.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Constellation operations means monitoring satellite health, predicting failures, and making intervention decisions for physical assets you can't physically reach. Your Trust essay's anomaly-review framework fits. Cummins/ZED Connect's predictive-asset-management design supports.
- Verdict: Ignore-until-trigger. Demand-type: constellation operations, anomaly review, or physical-infrastructure management design.
4. Zero-to-One AI Product Builds
Developer tools and infrastructure built from scratch — evaluation, tracing, code generation, data labeling. The thread across this cluster: making AI system internals (traces, evaluations, model behavior) legible to the humans who build and operate them. Queue up Agentic Labs (0→1 AI products), TinyFish (0→1 agentic platform in 3 months), and the Inference-Aware UX artifact.
31. Braintrust. AI evaluation and observability platform — traces, evals, dataset management. Series A, SF. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Unified evaluation across prompts, datasets, and model versions — the eval platform that treats AI quality as a continuous process.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Braintrust's core problem is your Inference-Aware UX artifact in action: a developer reviews a trace, identifies a failure, corrects the evaluation criteria, and needs to see how that correction changes the next run. TinyFish agent-trace experience is direct practitioner proof.
- Verdict: Ignore-until-trigger. Demand-type: human review of traces, evaluation design, or agent-release workflow.
32. Arize AI. ML observability and LLM tracing platform. Series B, SF. Andy Lu leads design. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Production ML observability — tracing model behavior from input to output with enough granularity to diagnose failures.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Arize's problem is tracing a failed output backward through data quality, latency, cost, and reasoning depth. Your Inference-Aware UX artifact was built for this exact diagnostic flow. TinyFish production traces are the practitioner proof.
- Verdict: Ignore-until-trigger. Demand-type: trace visualization, failure-attribution, or observability design.
33. Cognition. Devin — autonomous AI software engineer. Series B (~$2B valuation), SF. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Fully autonomous coding agent that takes a task description and produces working code — the most ambitious bet on AI-as-developer.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Devin's core interaction fits your Intent-Based Interaction artifact: a developer specifies what they want built, and needs to see the agent's plan, intervene mid-execution, and recover from mistakes without starting over.
- Verdict: Ignore-until-trigger. Demand-type: agent-planning visibility, mid-execution intervention, or autonomous-coding supervision design.
34. Cursor. AI-powered code editor with inline generation and chat. Series B, SF. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: AI code generation that lives inside the editor — co-authoring rather than delegating, which is why developers who tried Copilot switched.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cursor's problem is different from Cognition's: steering generation in progress rather than reviewing a completed plan. The interaction is closer to co-authoring than delegating — your Intent-Based Interaction artifact addresses this variation. Equinox+'s 0→MVP speed (3 months) shows craft-intensive product shipping.
- Verdict: Ignore-until-trigger. Demand-type: in-context generation steering, code-review surfaces, or AI-editor interaction design.
35. Snorkel AI. Data-centric AI platform — programmatic labeling, dataset management, evaluation. Series D, Redwood City. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Programmatic labeling that replaces manual annotation — the insight that data quality is a software problem, not a labor problem.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Snorkel's problem is dataset-quality decisions that determine model behavior downstream. Your Trust essay's evaluation framework fits. Thermo Fisher's 100% adoption in a quality-critical environment shows you can design evaluation workflows that experts actually trust.
- Verdict: Ignore-until-trigger. Demand-type: dataset-quality, evaluation-workflow, or labeling-review design.
36. Labelbox. Data labeling and AI evaluation platform. Series D, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Annotation workflow management at scale — the platform that enterprise ML teams use to manage human feedback quality.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Labelbox's core interaction: human evaluators provide feedback that changes model behavior, and the design must make the feedback loop visible. Your Human-Agent System Design artifact addresses this. Brand Pulse's real-time feedback architecture is the portfolio proof.
- Verdict: Ignore-until-trigger. Demand-type: evaluator queues, provenance, human-feedback quality, or annotation-workflow design.
37. Pinecone. Vector database for AI applications — retrieval, memory, agent data. Series C, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Purpose-built vector search that developers can deploy without managing infrastructure — the default vector database.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Retrieval quality and memory inspection become user-experience problems when agents use Pinecone as their knowledge layer — that's the Agent Infrastructure as UX artifact in practice. TinyFish production experience with retrieval architecture is the practitioner proof.
- Verdict: Ignore-until-trigger. Demand-type: retrieval-quality visualization, memory inspection, or agent-data governance design.
38. Together AI. Open-source model inference and fine-tuning platform. Series A, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Making open-source model inference fast and affordable — the alternative to closed-model APIs for teams that want control.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Together's core challenge is making model-selection and inference tradeoffs understandable to developers who aren't ML specialists. Your Inference-Aware UX artifact was designed for this.
- Verdict: Ignore-until-trigger. Demand-type: model-selection surfaces, inference-tradeoff visualization, or developer-facing AI-infrastructure design.
39. Anyscale. Distributed computing platform for AI workloads (Ray). Series C, SF. AI-native. C7/12 (AI 3, Stage 1, Design 1, Traj 2).
- Best at: Ray — the open-source distributed computing framework that most large-scale AI training runs depend on.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Distributed AI workloads need orchestration and debugging surfaces that make system behavior legible to developers. That's your Agent Infrastructure as UX artifact. TinyFish production-deployment experience adds the operational layer.
- Verdict: Ignore-until-trigger. Demand-type: workload orchestration, distributed-system debugging, or AI-infrastructure developer-experience design.
40. Sourcegraph. Code intelligence platform with Cody AI assistant. Late-stage, SF/remote. AI-native. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Code search and intelligence across massive codebases — the tool that makes enterprise-scale code navigable.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cody's problem: an AI assistant plans changes across a codebase, and the developer needs to review the plan and its implications before merge. Your Human-Agent System Design artifact addresses this. Trust essay's pre-execution review framework is the conceptual foundation.
- Verdict: Ignore-until-trigger. Demand-type: agent-planning review, codebase-wide change visualization, or AI-assistant interaction design.
5. Consumer AI at Scale
Consumer-facing AI for creative and content workflows — video, voice, documents, images. The running design problem: user control over generative output across iterations without losing intent or identity. Lead with Equinox+ (consumer 0→MVP), Allē (30M members, consumer scale), Trust essay (steering and consent), and the Intent-Based Interaction artifact.
41. Gamma. AI-powered presentation and document creation. Series B, SF. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: AI-native document creation that starts from intent rather than a blank page — presentations, docs, and websites generated from a description.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Gamma's core problem: users refine generated documents iteratively, and the system must preserve structural intent across revisions. Your Intent-Based Interaction artifact addresses this. Allē's 30M-member dual-surface redesign shows consumer-scale design.
- Verdict: Ignore-until-trigger. Demand-type: iterative-generation steering, document-structure preservation, or consumer AI-creation design.
42. Pika. AI video generation and editing. Series B, SF. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Video generation that feels like directing — scene-level control over generated video rather than prompt-and-pray.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Pika's problem is steering generated video across iterations without restarting the creative process. Your Intent-Based Interaction artifact applies differently here than at Gamma — the interaction model is closer to directing than editing. Equinox+'s 0→MVP in 3 months shows craft-intensive consumer product speed.
- Verdict: Ignore-until-trigger. Demand-type: video-generation steering, iterative creative control, or consumer AI-media design.
43. ElevenLabs. AI voice synthesis and cloning platform. Series C, SF/London. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Voice quality that crosses the uncanny valley — the first AI voice platform where output sounds genuinely human.
- Signals: No recent signal. Last assessed Issue #10.
- Match: ElevenLabs' hardest problem is voice identity: cloning requires consent, disclosure, and provenance that users trust. Trust design at the identity layer, and your Trust essay covers it directly.
- Verdict: Ignore-until-trigger. Demand-type: voice-identity consent, provenance, disclosure, or trust-layer design.
44. Descript. AI-powered video and audio editing platform. Series D, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Editing video by editing text — the transcript-first editing model that made video production accessible to non-editors.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Descript's review problem: AI edits across transcript, audio, and video simultaneously, and the professional user needs to review changes across all three surfaces before publishing. Your Human-Agent System Design artifact fits.
- Verdict: Ignore-until-trigger. Demand-type: multi-surface AI-edit review, professional-production, or transcript-based editing design.
45. Captions. AI video creation — avatars, translation, editing. Series C, NYC. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay eligibility.
- Best at: AI-generated video avatars and automatic translation — making one creator's content work in any language with their face.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Captions' avatar problem: generating a video avatar of a real person requires consent, control, and the ability to revoke. Adjacent to ElevenLabs but applied to visual identity rather than voice. Your Trust essay's identity-consent framework covers both.
- Verdict: Ignore-until-trigger. Demand-type: avatar control, identity consent, multi-stage generation, or localized-video design.
46. Typeface. Enterprise generative AI for brand content. Series C, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Brand-safe enterprise content generation — the guardrails that let marketing teams use AI without going off-brand.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Typeface's core problem: enterprise content generation must stay within brand constraints, and approval workflows must be lightweight enough that teams actually use them. Your Trust essay's governance framework fits. Allē's brand-constrained dual-surface design is the proof.
- Verdict: Ignore-until-trigger. Demand-type: brand-governed generation, enterprise-approval workflow, or content-governance design.
47. HeyGen. AI video creation — avatars, translation, enterprise video production. Series B, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Enterprise video production at scale — multiple team members creating avatar content with translation, permissions, and brand control.
- Signals: No recent signal. Last assessed Issue #10.
- Match: The distinction from Captions is team production: multiple people creating avatar content for an enterprise, which requires permissions and approval. Alibaba's multi-stakeholder platform design separates you from candidates who've only designed for individual creators.
- Verdict: Ignore-until-trigger. Demand-type: avatar control, team production, translation, or enterprise-video consent design.
48. Recraft. AI image generation focused on production-ready design assets. Series A, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Production-ready vector and image output — the AI image tool that designers use because it understands what "production-ready" means.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Recraft's problem: designers need precise style control over generated output — production-ready vector assets that meet professional standards. Your Intent-Based Interaction artifact addresses this. The differentiator is that you understand what "production-ready" means to a designer because you are one.
- Verdict: Ignore-until-trigger. Demand-type: style-control surfaces, production-asset generation, or professional design-tool AI design.
49. Krea. Real-time multimodal AI generation — images, video, live canvas. Series A, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Real-time generation — the canvas updates as you work, so creation and generation happen simultaneously rather than sequentially.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Krea's problem is unique in this cluster: generation happens in real time, so the user's intent evolves as they see results. The interaction model is closer to improvisation than specification — your Intent-Based Interaction artifact applies differently here. Equinox+'s rapid MVP shows you can design for emergent user behavior.
- Verdict: Ignore-until-trigger. Demand-type: live-generation interaction, real-time creative-steering, or multimodal-canvas design.
50. Jasper. AI marketing content platform for enterprise teams. Series A (post-pivot), SF. AI-native. C7/12 (AI 3, Stage 1, Design 1, Traj 2).
- Best at: Enterprise marketing content at scale — the platform that marketing teams adopted first, now rebuilding around agents after a pivot.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Jasper's problem: campaign-generation agents that produce content within brand constraints and require approval before publication. Allē's marketing-platform design (3.2× redemption, $42 CAC) is the proof. Trust essay's governance framework extends it.
- Verdict: Ignore-until-trigger. Demand-type: brand-constrained generation, campaign-approval, or marketing-agent design.
6. Regulated and Healthcare Workflows
Healthcare and regulated domains where AI assists clinical or administrative decisions, all constrained by escalation boundaries, clinical evidence requirements, and compliance — a wrong output means patient harm or regulatory violation. Your proof: Thermo Fisher (regulated platform, pharma, 100% adoption), Red Cross (high-stakes disbursement — disaster relief, not healthcare), and Trust essay (escalation ladder).
51. Hippocratic AI. Patient-facing AI agents for healthcare — staffing, triage, follow-up. Series B, Palo Alto. Healthcare/Regulated. C9/12 (AI 3, Stage 2, Design 2, Traj 2).
- Best at: Patient-facing AI agents with safety guardrails — the company betting that AI can handle non-diagnostic patient interactions safely.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Hippocratic's specific problem: the agent is talking to a patient, detects something outside its scope, and must hand off to a human clinician without alarming the patient or losing context. Your Trust essay's escalation ladder addresses this precisely. Thermo Fisher's 100% adoption in pharma shows regulated-user trust.
- Verdict: Ignore-until-trigger. Demand-type: patient-agent escalation, clinical handoff, or healthcare-agent boundary design.
52. AKASA. AI for healthcare revenue-cycle management — coding, billing, claims. Series B, SF. Healthcare/Regulated. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Medical coding automation with human-in-the-loop — the AI that handles routine claims so coders can focus on complex cases.
- Signals: No recent signal. Last assessed Issue #10.
- Match: AKASA's core interaction: an AI codes a medical claim, and a human coder reviews exceptions before release. Your Trust essay's Output Review pattern fits. Thermo Fisher's regulated-workflow design ($20M+ margin) shows you can build review workflows in compliance-critical environments.
- Verdict: Ignore-until-trigger. Demand-type: medical-coding exceptions, human-release workflows, or revenue-cycle review design.
53. Transcarent. AI-powered health and care navigation platform. Series D, SF. Healthcare/Regulated. C8/12 (AI 2, Stage 2, Design 2, Traj 2).
- Best at: Employer-sponsored care navigation that uses AI to guide members to the right care at the right cost.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Transcarent's problem: AI-guided care navigation must show patients why a recommendation was made and when to escalate to a human navigator. Your Trust essay's transparency framework covers this. Red Cross's national-scale platform shows design for diverse user populations under stress.
- Verdict: Ignore-until-trigger. Demand-type: AI-guided care navigation, escalation transparency, or patient-decision design.
54. Innovaccer. Healthcare data platform — unified patient records, AI-powered analytics. Series E, SF. Healthcare/Regulated. C8/12 (AI 2, Stage 2, Design 2, Traj 2).
- Best at: Unified patient data across fragmented provider systems — the integration layer that makes healthcare data usable.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Innovaccer's problem is platform coherence: multiple provider types need consistent decision support across fragmented data. Alibaba's platform redesign ($50B+ GMV, cross-functional sprints) is the proof. Thermo Fisher's pharma-partner adoption adds regulated-domain credibility.
- Verdict: Ignore-until-trigger. Demand-type: provider decision-support, patient-data governance, or healthcare-platform design.
55. OpenEvidence. AI-powered clinical evidence synthesis for physicians. Growth-stage, Boston. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Clinical evidence synthesis that physicians actually trust — answers grounded in peer-reviewed literature with inspectable citations.
- Signals: No recent signal. Last assessed Issue #10.
- Match: OpenEvidence's core interaction: a physician reads an AI-synthesized answer and needs to inspect and challenge the cited sources before acting on it clinically. Carrier IQ's provenance design is the direct match.
- Verdict: Ignore-until-trigger. Demand-type: citation inspection, clinical-evidence challenge, or synthesized-answer design.
56. Regard. AI diagnostic and clinical-reasoning assistant for physicians. Growth-stage, NYC. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Automated clinical reasoning that surfaces diagnoses physicians might miss — the AI second opinion embedded in the EHR.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Regard's core problem: a physician sees an AI diagnostic suggestion and needs to understand the reasoning, agree or disagree, and document the override. Your Trust essay's Override pattern covers this. Thermo Fisher's regulated-user adoption shows you can design for expert users who maintain professional judgment.
- Verdict: Ignore-until-trigger. Demand-type: physician review, diagnostic override, or clinical-reasoning design.
57. Notable. AI-powered patient intake and clinical workflow automation. Series B, SF. Healthcare/Regulated. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Patient intake automation that reduces staff burden — the AI that handles forms, scheduling, and pre-visit workflows.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Notable's problem: an AI handles patient intake, and the handoff to clinical staff must preserve context without requiring re-asked questions. Your Trust essay's handoff framework addresses this. Red Cross's multi-system consolidation shows handoff design across fragmented workflows.
- Verdict: Ignore-until-trigger. Demand-type: patient-intake automation, agent-to-staff handoff, or clinical-workflow design.
58. Cohere Health. AI-powered utilization management and prior authorization. Series C, Boston. Healthcare/Regulated. C7/12 (AI 2, Stage 2, Design 1, Traj 2). Later research suggested a higher assessment — reconcile on trigger. Flag: verify Bay/remote eligibility.
- Best at: Prior authorization automation that reduces approval delays — the platform payers use to process auth requests faster.
- Signals: No recent signal. Last assessed Issue #10. Score may need upward revision per Issue #11.
- Match: Cohere Health's core interaction: an AI makes a prior-authorization determination, and a clinician reviews the evidence before release. Your Trust essay's determination-review framework fits. Thermo Fisher's 100% adoption in a review-intensive regulated environment is the proof.
- Verdict: Ignore-until-trigger. Demand-type: determination review, clinician handoff, or utilization-management design.
59. Cedar. AI-powered patient billing and financial engagement. Series D, NYC. Healthcare/Regulated. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Patient billing experience that actually gets paid — the platform that makes medical bills understandable and payable.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cedar's problem: a billing agent takes a financial action on a patient's account, and the patient needs to understand what happened and recover from errors. Your Trust essay's reversibility framework addresses this. Allē's consumer financial platform (30M members, $42 CAC) shows financial interactions at consumer scale.
- Verdict: Ignore-until-trigger. Demand-type: billing-agent transparency, patient financial recovery, or healthcare-billing design.
60. Tennr. AI-powered referral management for healthcare. Series A, NYC. Healthcare/Regulated. C7/12 (AI 3, Stage 1, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Referral processing automation — extracting, routing, and completing referrals that otherwise sit in fax queues.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Tennr's specific problem: a referral is missing evidence, and the system must flag the gap and enable recovery before care is delayed. Your Trust essay's exception-handling framework fits. Red Cross's high-consequence platform shows time-sensitive exception handling.
- Verdict: Ignore-until-trigger. Demand-type: referral exceptions, missing-evidence recovery, or care-delay prevention design.
7. Infrastructure and Operational Platforms
Operational infrastructure across field service, manufacturing, maintenance, HR, and energy. The recurring design problem: AI predictions and recommendations that operators who aren't data scientists can actually act on. Start with Thermo Fisher (operational infrastructure, 100% adoption), Cummins/ZED Connect (IoT fleet management — supporting proof, not lead), and Alibaba (enterprise platform at scale).
61. ServiceTitan. Operating platform for home-service businesses — HVAC, plumbing, electrical. Recently IPO'd, Glendale. Enterprise platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility — HQ is LA area.
- Best at: The operating system for trades businesses — scheduling, dispatch, invoicing, and marketing in one platform.
- Signals: No recent signal. Last assessed Issue #10.
- Match: ServiceTitan's operational problem: dispatchers, technicians, and office staff all need different views of the same AI-optimized schedule. Cummins/ZED Connect's dual-surface fleet management addresses this. Alibaba's multi-stakeholder platform design adds scale proof.
- Verdict: Ignore-until-trigger. Demand-type: AI dispatch, cross-role operational control, or field-service platform design.
62. MaintainX. Connected-worker platform for industrial maintenance and operations. Series C, SF. Joshua Chauvin leads design. Growth-stage platform. C8/12 (AI 2, Stage 2, Design 2, Traj 2).
- Best at: Mobile-first maintenance management that frontline workers actually use — work orders, procedures, and asset tracking on the factory floor.
- Signals: No recent signal. Last assessed Issue #10.
- Match: MaintainX's design problem: an AI predicts a maintenance need, generates a work order, and the technician needs to see the evidence behind the prediction before acting. Cummins/ZED Connect's predictive-asset-management design is the direct match. Thermo Fisher's 100% adoption shows design for frontline workers.
- Verdict: Ignore-until-trigger. Demand-type: AI-generated work orders, predictive-maintenance evidence, or connected-worker design.
63. Motive. Fleet management and safety platform — AI dashcams, ELD, fleet tracking. Late-stage, SF. Enterprise platform. C8/12 (AI 2, Stage 2, Design 2, Traj 2).
- Best at: AI-powered fleet safety that uses dashcam footage to detect and coach risky driving behavior in real time.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cummins/ZED Connect (fleet management, driver tracking) is the direct domain match. The specific design problem: a safety event is flagged, and the fleet manager needs to understand system confidence before intervening with the driver. Your Trust essay's confidence framework addresses this.
- Verdict: Ignore-until-trigger. Demand-type: driver-safety event confidence, fleet intervention, or safety-alert design.
64. Eightfold AI. AI-powered talent intelligence platform. Late-stage, Santa Clara. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Deep-learning talent matching that goes beyond keyword search — the platform that matches candidates to roles based on skills and potential.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Eightfold's hardest problem: an AI matches a candidate to a role, and the recruiter needs to understand why — and the candidate needs to understand why they were rejected. Consequential decisions with explanation obligations. Your Trust essay's explanation framework fits.
- Verdict: Ignore-until-trigger. Demand-type: consequential-matching explanation, human review of AI talent decisions, or hiring-intelligence design.
65. Procore. Construction management platform with AI capabilities. Public, Carpinteria/Austin. Enterprise platform. C6/12 (AI 1, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility — HQ is Central Coast CA.
- Best at: Construction project management that connects the office to the field — the platform general contractors standardize on.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Procore's problem: contractors, finance, and field operations all interact with the same project data, and AI project controls must be legible across all three audiences. Alibaba's multi-stakeholder platform redesign is the proof.
- Verdict: Ignore-until-trigger. Demand-type: AI project controls, cross-stakeholder construction, or field-operations design.
66. Juniper Square. Investment management platform for private markets. Series D, SF. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2).
- Best at: Fund administration and investor reporting for private capital — the platform that replaced spreadsheets for GP-LP operations.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Juniper Square's problem is fund-accounting review and AI document workflows in a compliance-sensitive environment. Thermo Fisher's regulated-platform design ($20M+ margin, pharma partners) is the proof.
- Verdict: Ignore-until-trigger. Demand-type: fund-accounting review, AI document workflows, or financial-operations design.
67. Tulip. Manufacturing operations platform — frontline apps, IoT, quality management. Series D, Somerville MA. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: No-code app building for factory floors — letting manufacturing engineers create frontline apps without IT involvement.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Tulip's problem: an operator on the factory floor encounters an AI-assisted quality decision and needs to release or reject based on evidence they can understand. Cummins/ZED Connect's IoT-connected operations design addresses this. Thermo Fisher's 100% adoption shows design for frontline users in manufacturing.
- Verdict: Ignore-until-trigger. Demand-type: manufacturing exceptions, operator-release, or AI-assisted quality-decision design.
68. Watershed. Climate and sustainability data platform. Series C, SF. Growth-stage platform. C8/12 (AI 1, Stage 3, Design 2, Traj 2). Current Staff role is London-only — geographic mismatch.
- Best at: Enterprise carbon accounting that holds up to audit — the platform companies use to measure, report, and reduce emissions.
- Signals: Staff Product Designer role live but London-only. No Bay-eligible qualifying role.
- Match: Watershed's problem: emissions data must be traceable, auditable, and defensible to regulators. Thermo Fisher's audit-ready regulated-platform design is the proof.
- Verdict: Ignore-until-trigger. Demand-type: Bay-eligible or US-remote Staff+ role, or any audit-ready reporting/sustainability-data design role.
69. Crusoe. AI-focused cloud infrastructure powered by stranded energy. Series C, SF. Growth-stage platform. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: AI compute powered by otherwise-wasted energy — the cloud provider that turns stranded natural gas into GPU capacity.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Crusoe's problem is capacity decisions and energy-source management for AI workloads. Cummins/ZED Connect's operational-monitoring design fits.
- Verdict: Ignore-until-trigger. Demand-type: energy-aware operations, capacity-decision, or cloud-infrastructure design.
70. Arcadia. Utility data platform — energy analytics, grid intelligence. Series D, DC. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Utility data access at scale — the API layer that makes fragmented utility data usable for energy applications.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Arcadia's problem: utility data has permission constraints, and analytics must respect those constraints while remaining useful. Thermo Fisher's regulated-data-platform design is the proof.
- Verdict: Ignore-until-trigger. Demand-type: utility-data permissions, energy-analytics, or regulated-data-platform design.
8. Marketplace and Multi-Sided Platforms
Multiple parties, supply chain, logistics, commerce. The design problem across this cluster is cross-party visibility, exception handling, and trust when multiple stakeholders see different views of the same transaction. Lead with Alibaba (B2B marketplace, $50B+ GMV, +20% transactions), Allē (dual-surface platform), and Trust essay (multi-party trust).
71. Flexport. Global freight forwarding and supply-chain platform. Late-stage, SF. Enterprise platform. C8/12 (AI 1, Stage 2, Design 3, Traj 2).
- Best at: Full-stack freight forwarding with software visibility — the platform that made global logistics trackable for mid-market shippers.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Flexport's multi-party platform problem is Alibaba territory. Your B2B platform redesign ($50B+ GMV, cross-functional sprints across homepage, search, and PDP) is the direct proof. The +20% daily transactions shows you can redesign a complex marketplace without disrupting active trade.
- Verdict: Ignore-until-trigger. Demand-type: customs-evidence design, shipment-planning, or multi-party logistics-platform design.
72. Altana. AI-powered supply-chain intelligence and compliance platform. Series C, NYC. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Supply-chain knowledge graph that maps entity relationships across global trade — the compliance intelligence layer.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Altana's problem: an entity in the supply chain is flagged, and the compliance analyst needs to inspect the evidence graph before acting. Your Trust essay's evidence-inspection framework addresses this. Carrier IQ's provenance design is the portfolio proof.
- Verdict: Ignore-until-trigger. Demand-type: entity-evidence inspection, supply-chain compliance, or knowledge-graph design.
73. FourKites. Real-time supply-chain visibility platform. Late-stage, Chicago. Enterprise platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Real-time shipment visibility across modes — the platform shippers use to track freight regardless of carrier.
- Signals: No recent signal. Last assessed Issue #10.
- Match: FourKites' problem: a shipment delay is predicted, and multiple parties need to see the prediction, understand the evidence, and coordinate a response. Cummins/ZED Connect's predictive-service design fits. Alibaba's multi-stakeholder platform adds scale proof.
- Verdict: Ignore-until-trigger. Demand-type: predictive-exception design, cross-party resolution, or supply-chain visibility design.
74. project44. Supply-chain visibility and intelligence platform. Late-stage, Chicago. Enterprise platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Carrier connectivity breadth — more carrier integrations than competitors, which means more complete visibility data.
- Signals: No recent signal. Last assessed Issue #10.
- Match: project44's problem: predicted delays require resolution across parties with different information and different incentives. Alibaba's cross-functional platform design is the proof.
- Verdict: Ignore-until-trigger. Demand-type: cross-party delay resolution, predictive logistics, or multi-stakeholder visibility design.
75. Platform Science. Connected-vehicle platform for commercial fleets. Growth-stage, San Diego. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Open connected-vehicle platform that lets fleets run multiple apps on one hardware device, replacing dedicated hardware per vendor.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cummins/ZED Connect is the exact match here — connected-vehicle IoT, driver workflows, fleet management. The dual-surface design experience (mobile for drivers, web for fleet managers) is the proof.
- Verdict: Ignore-until-trigger. Demand-type: driver-workflow, fleet-permissions, or connected-vehicle ecosystem design.
76. Augury. AI-powered predictive maintenance for industrial machines. Series D, NYC. AI-native. C7/12 (AI 3, Stage 1, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Machine-health diagnostics from vibration and temperature data — predictive maintenance that catches failures before they happen.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Augury's problem: an AI predicts a machine failure, and the maintenance team needs to see the evidence (vibration data, temperature trends) before scheduling intervention. Cummins/ZED Connect's predictive-asset-management design is the match.
- Verdict: Ignore-until-trigger. Demand-type: predictive-maintenance evidence, intervention-recommendation, or industrial-IoT design.
77. Stord. Cloud supply-chain platform — fulfillment, warehousing, transportation. Series D, Atlanta. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Unified fulfillment across owned and partner warehouses — the platform that lets brands scale logistics without building their own network.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Stord's multi-party problem: merchants and warehouse operators see different views of the same inventory, and exceptions need resolution across both. Alibaba's B2B platform redesign is the proof. Allē's dual-surface design adds multi-audience platform evidence.
- Verdict: Ignore-until-trigger. Demand-type: inventory-intelligence, fulfillment-exception, or multi-party warehouse design.
78. Fleetio. Fleet management software — maintenance, fuel, inspections. Growth-stage, Birmingham AL. Growth-stage platform. C6/12 (AI 1, Stage 1, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Fleet maintenance management that's simple enough for small and mid-size fleets — the Fleetio-to-enterprise pipeline.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cummins/ZED Connect is the strongest match in this cluster — fleet management, predictive maintenance, dual-surface design for drivers and fleet managers. The domain overlap is near-total.
- Verdict: Ignore-until-trigger. Demand-type: fleet-maintenance prediction, technician-workflow, or fleet-management design.
79. Constructor. AI-powered product search and discovery for e-commerce. Series A, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: AI product search that optimizes for revenue rather than relevance — the search engine merchants use because it sells more.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Constructor's core problem: merchants need to understand and control how AI ranks and recommends their products. Alibaba's search redesign (+20% daily transactions) is the direct proof.
- Verdict: Ignore-until-trigger. Demand-type: merchant-control surfaces, AI-recommendation explanation, or e-commerce search design.
80. Rokt. AI-powered post-purchase marketing and transaction optimization. Late-stage, NYC. AI-native. C7/12 (AI 2, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Post-purchase transaction moment — the platform that monetizes the confirmation page with relevant offers rather than generic ads.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Rokt's problem: merchants need to control and understand AI-generated post-purchase offers without disrupting the transaction experience. Allē's marketing-platform design (3.2× redemption, $42 CAC) fits.
- Verdict: Ignore-until-trigger. Demand-type: merchant-control, post-purchase AI, or transaction-optimization design.
9. Fintech and Financial Automation
Automated financial decisions — risk scoring, identity verification, fraud detection, underwriting. The regulatory thread: explainability, human review, and recovery from adverse automated decisions where the person affected has legal rights to understand why. Reach for Allē (financial transactions at scale), Trust essay (reversibility and explanation), and Thermo Fisher (regulated-platform adoption).
81. Oscilar. AI-powered risk decisioning platform for fintechs. Series A, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Real-time risk decisioning that fintechs can deploy without building their own ML pipeline.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Oscilar's core problem: a risk decision is made, and the affected person (or the compliance team) needs to understand why. Your Trust essay's explanation framework addresses this. Carrier IQ's provenance design shows you can make automated decisions inspectable.
- Verdict: Ignore-until-trigger. Demand-type: risk-decision explanation, investigation-evidence, or adverse-action design.
82. Sardine. AI fraud and compliance platform for fintechs. Series C, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Device intelligence and behavioral biometrics for fraud detection — seeing fraud signals that transaction data alone misses.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Sardine's problem: a transaction is blocked by AI, and a human analyst reviews the case evidence before releasing or confirming the block. Your Trust essay's human-release framework fits. Red Cross's high-consequence decision design shows time-sensitive release decisions.
- Verdict: Ignore-until-trigger. Demand-type: transaction-monitoring cases, human-release workflows, or fraud-investigation design.
83. Alloy. Identity decisioning platform — KYC, KYB, fraud prevention. Series C, NYC. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Orchestrating identity verification across multiple data sources in one decisioning flow — the platform banks use to build KYC without vendor lock-in.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Alloy's problem: compliance teams build identity-verification rules, and the system must make the rules' effects visible before they go live. Your Trust essay's policy-builder framework addresses this. Thermo Fisher's 100% adoption in pharma shows you can design policy tools that compliance users trust.
- Verdict: Ignore-until-trigger. Demand-type: identity-decision explanation, policy-builder, or manual-review-queue design.
84. Persona. Identity verification and fraud prevention platform. Series C, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Identity verification UX that converts — the platform where the verification step doesn't kill the signup funnel.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Persona's specific problem: a user fails an identity check, and the recovery path must be clear, trustworthy, and fast. Your Trust essay's reversibility framework fits. Allē's consumer-scale platform (30M members) shows identity-adjacent experiences at scale.
- Verdict: Ignore-until-trigger. Demand-type: verification-policy design, user-recovery, or identity-check failure-handling design.
85. Middesk. Business identity verification for financial services. Series B, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Business identity verification that automates what underwriters used to do manually — pulling registrations, liens, and litigation records.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Middesk's problem: underwriters need to see the evidence behind a business-identity verification before making a lending decision. Carrier IQ's provenance design is the direct match.
- Verdict: Ignore-until-trigger. Demand-type: business-identity evidence, underwriting-evidence, or compliance-verification design.
86. Sixfold. AI-powered insurance underwriting platform. Series A, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Automated commercial insurance underwriting — the platform that turns submission documents into risk assessments.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Carrier IQ's InsurTech quote-automation design is the direct domain match. Your Trust essay's override framework addresses what happens when an underwriter disagrees with the AI's risk assessment and needs to document why.
- Verdict: Ignore-until-trigger. Demand-type: underwriter inspection, AI-evidence override, or insurance-underwriting design.
87. Coalition. Cyber insurance and security platform. Series F, SF. AI-native. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Active insurance — combining cyber insurance with continuous security monitoring, so the insurer also prevents the claims.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Carrier IQ gives you insurance-domain credibility, but the differentiator here is lifecycle continuity: risk explanation through incident detection through claim resolution. Your Trust essay's evidence-chain framework addresses that continuity.
- Verdict: Ignore-until-trigger. Demand-type: risk-explanation, incident-to-claim continuity, or cyber-insurance design.
88. Sift. AI fraud detection and prevention platform. Late-stage, SF. AI-native. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Cross-platform fraud signals — the network effect of seeing fraud patterns across thousands of merchants.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Sift's specific problem: a legitimate user is flagged as fraudulent, and the recovery path must be fast, clear, and preserve trust. Your Trust essay's false-positive recovery framework fits. Allē's consumer-scale platform shows trust-sensitive experiences at scale.
- Verdict: Ignore-until-trigger. Demand-type: false-positive recovery, fraud-evidence explanation, or trust-recovery design.
89. Highnote. Modern card-issuing and payment platform. Series B, SF. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2).
- Best at: Modern card-issuing infrastructure — the API-first platform that lets fintechs launch card programs without legacy processor dependencies.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Highnote's multi-party problem: card issuers, cardholders, and merchants all interact with the same financial infrastructure. Allē's dual-surface financial platform (consumer + provider) is the proof.
- Verdict: Ignore-until-trigger. Demand-type: card-control design, multi-party financial operations, or payment-platform design.
90. Ramp. Corporate spend management platform with AI-powered automation. Late-stage (~$13B valuation), NYC. Growth-stage platform. C10/12 (AI 2, Stage 3, Design 3, Traj 2). Current Director role is NYC-hybrid only — geographic mismatch. Held here rather than on the active list because the existing role requires three days a week in NYC.
- Best at: Corporate card and spend management that saves finance teams time — the fastest-growing spend platform because the product actually reduces work.
- Signals: Director, Product Design live at $320K–$440K plus equity, but NYC-hybrid only. No Bay-eligible qualifying role.
- Match: Ramp's hardest problem: an AI agent executes a financial action (expense categorization, policy enforcement, payment), and the action must be reversible if wrong. Your Trust essay's reversibility framework addresses this. TinyFish's agentic-platform experience adds the operational layer. Alibaba's +20% transactions shows platform-scale financial design.
- Verdict: Ignore-until-trigger. Demand-type: Bay-eligible or US-remote Director+ or Staff+ role | agentic financial operations, reversible high-consequence actions, or AI-spend-management design.
10. Hardware-Software Integration and Spatial Computing
Software controlling physical systems — autonomous vehicles, drones, robots, energy infrastructure. The hard part: communicating system intent and capability to humans when the system acts autonomously in the physical world. Pull Red Cross (high-consequence operations) and Trust essay (delegation and override). Cummins/ZED Connect has the strongest domain relevance in this cluster — use it as supporting proof, leading with Trust essay or Red Cross for the conceptual framework.
91. Zoox. Autonomous ride-hailing vehicles — no steering wheel, no pedals. Subsidiary of Amazon, Foster City. AI-native. C9/12 (AI 3, Stage 2, Design 2, Traj 2). No qualifying design posting in current Lever inventory.
- Best at: Purpose-built autonomous vehicles designed from scratch for riders, not retrofitted cars — the only AV company that started with the vehicle form factor.
- Signals: No qualifying design posting found. Extensive autonomy, vehicle, and operations hiring active.
- Match: Zoox's problem: the rider needs to understand vehicle intent and capability, while fleet operations needs predictive maintenance and anomaly review. Cummins/ZED Connect's dual-surface fleet management (consumer-facing IoT + predictive asset management) is the strongest domain match in this cluster.
- Verdict: Ignore-until-trigger. Demand-type: rider-experience, autonomy-supervision, or Staff+ vehicle-intent design.
92. Skydio. Autonomous drones for enterprise and defense — inspection, mapping, security. Late-stage, San Mateo. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Autonomous obstacle avoidance that lets non-pilots fly complex inspection missions — the drone that flies itself.
- Signals: No recent signal. Last assessed Issue #10.
- Match: When should an operator intervene in an autonomous mission? Your Trust essay's delegation ladder addresses this. Cummins/ZED Connect's fleet-management design supports with mission planning, remote supervision, and incident review across drone fleets.
- Verdict: Ignore-until-trigger. Demand-type: mission-planning, remote-supervision, or autonomous-drone-fleet design.
93. Zipline. Autonomous drone delivery for medical supplies. Late-stage, SF. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2).
- Best at: Autonomous medical delivery at scale — the only drone delivery company with thousands of daily commercial flights in production.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Red Cross's medical-supply disbursement platform ($847K, national deployment) is the domain match: medical delivery where timing matters and failure has health consequences. Your Trust essay's state-visibility framework addresses delivery tracking for healthcare recipients.
- Verdict: Ignore-until-trigger. Demand-type: medical-delivery state, remote intervention, or fleet-exception design.
94. Gecko Robotics. Inspection robots for industrial infrastructure — power plants, refineries, tanks. Series C, Pittsburgh. AI-native. C8/12 (AI 3, Stage 2, Design 1, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Robot-collected inspection data that creates digital twins of industrial infrastructure — seeing inside assets that humans can't safely access.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Gecko's problem: inspection data from robots generates maintenance predictions, and the engineer needs to see the evidence before scheduling work on critical infrastructure. Your Trust essay's evidence-inspection framework fits. Cummins/ZED Connect and Thermo Fisher support for predictive asset management and regulated-environment adoption.
- Verdict: Ignore-until-trigger. Demand-type: inspection-evidence, predictive-maintenance, or industrial-infrastructure design.
95. Verkada. AI-powered physical security — cameras, access control, environmental sensors. Late-stage, San Mateo. AI-native. C7/12 (AI 2, Stage 2, Design 1, Traj 2).
- Best at: Cloud-managed physical security that unifies cameras, access, and sensors in one dashboard — the platform that replaced DVRs.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Verkada's problem: an AI-generated security alert fires, and the operator needs to understand system confidence before dispatching a response. Your Trust essay's confidence framework fits. Brand Pulse's real-time monitoring architecture is the portfolio proof.
- Verdict: Ignore-until-trigger. Demand-type: AI-alert confidence, policy-control, or physical-security design.
96. Span. Smart electrical panel and home energy management. Series B, SF. Hardware-software integration. C8/12 (AI 1, Stage 2, Design 3, Traj 2).
- Best at: The smart electrical panel — the hardware that makes home energy visible and controllable at the circuit level.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Span's problem: a homeowner needs to understand energy flow across multiple sources and recover from hardware or grid failures. Cummins/ZED Connect's IoT-connected operations design fits — the dual-surface (panel hardware + app software) experience is the specific differentiator.
- Verdict: Ignore-until-trigger. Demand-type: energy-orchestration, hardware-software integration, or home-energy design.
97. Form Energy. Multi-day energy storage systems for the grid. Growth-stage, Somerville MA. Hardware-software integration. C7/12 (AI 1, Stage 1, Design 2, Traj 3). Flag: verify Bay/remote eligibility.
- Best at: Iron-air battery technology that stores energy for 100+ hours — the storage duration that makes renewables viable as baseload power.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Form Energy's problem: multi-day storage operations require field decisions that affect grid reliability. Your Trust essay's operational-decision framework fits. Cummins/ZED Connect's operational-monitoring design supports.
- Verdict: Ignore-until-trigger. Demand-type: storage-operations, grid-management, or field-decision design.
98. Redwood Materials. Battery recycling and materials recovery for the circular economy. Growth-stage, Carson City NV. Hardware-software integration. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Closed-loop battery recycling at commercial scale — recovering cathode and anode materials from end-of-life batteries.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Redwood's problem is materials traceability: trust across multiple parties in the recycling chain. Thermo Fisher's supply-chain platform design (6 pharma partners, 100% adoption) is the proof.
- Verdict: Ignore-until-trigger. Demand-type: materials-traceability, manufacturing-operations, or circular-economy design.
99. EquipmentShare. Construction equipment rental and fleet management platform. Late-stage, Columbia MO. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Flag: verify Bay/remote eligibility.
- Best at: Equipment rental with built-in telematics — the platform that combines rental, tracking, and maintenance for construction fleets.
- Signals: No recent signal. Last assessed Issue #10.
- Match: Cummins/ZED Connect is the direct match — fleet management, predictive service, dual-surface design for operators and fleet managers. Construction-equipment domain is adjacent to commercial-vehicle fleet management.
- Verdict: Ignore-until-trigger. Demand-type: fleet-intelligence, predictive-service, or construction-equipment design.
100. Aurora Solar. Solar design and sales platform for installers. Late-stage, SF. Growth-stage platform. C7/12 (AI 1, Stage 2, Design 2, Traj 2). Prior design-leadership search was Canada-specific — trigger must be US-eligible.
- Best at: AI-powered solar system design that generates permit-ready plans from a satellite image — the tool that makes solar proposals fast.
- Signals: No recent signal. Prior Director-level design search was Canada-only.
- Match: Aurora Solar's problem: installers adopt the platform only if the design workflow saves time over their current process. Thermo Fisher's partner-adoption design (6 partners, 100% adoption) is the proof.
- Verdict: Ignore-until-trigger. Demand-type: US-eligible Staff+ or Director+ role | spatial-design, installer-workflow, or solar-platform design.
Check these triggers weekly against your LinkedIn alerts and ATS notifications. When one fires, score the actual role against the role rubric before acting — a trigger is a reason to look, not a reason to apply. The entry gives you the company score, portfolio match, and tier archetype. Read the tier playbook, then move.
- Glean's Head of Product Design: This Mountain View/SF hybrid role at $250K-$300K surfaced during dormant-list verification and needs full rubric scoring, decision-maker identification, and portfolio routing in the next active-signal cycle.
- Ramp's geographic constraint: The Director, Product Design at $320K-$440K requires three days a week in NYC — worth monitoring whether Ramp adds a Bay-eligible copy or relaxes the hybrid requirement as the search ages.
- Amplitude's unresolved search state: The Head of Product Design role remained visible on LinkedIn after a reported September 17 application deadline and is now past the 43-day decay threshold — a repost or new requisition number would be the signal to re-engage.
- Affirm's removed agentic role: The Staff Product Designer, Agentic Experiences was recorded as removed on September 18 — if Affirm reposts with a new requisition or adjacent agentic-finance title, it would qualify immediately given the agent-trust demand family match.